| Author | Thread |
|
|
02/25/2003 08:28:14 PM · #1 |
| I was looking through the scores of the winning photographs in the archives, and noticed that they all scored around a 7.45, which makes the score for a decent photograph around 3.72. Realizing this has made me feel better about the scores I received for my first few submissions, and has really given me some encouragement to get better. |
|
|
|
02/25/2003 08:36:21 PM · #2 |
| I hate to burst your bubble, but statistics don't work that way. Its a normal distribution. The average is probably in the 4.5 to 5.5 range. Take that plus or minus two times the standard deviation and you'll probably get close to the top and bottom scores - if i remember this correctly from college. |
|
|
|
02/25/2003 08:40:08 PM · #3 |
just looking at the last 10 challenges, the average winning score was 7.36 while the average last-place entry scored 3.13 --- so the mid-range score was about 5.25
in order to get more accurate results, ALL scores must be averaged, not just the highs & lows
|
|
|
|
02/25/2003 09:12:19 PM · #4 |
OK, you guys piqued the interest of the statistician in me again...
I haven't actually run statistical tests on the data, but the distributions of the voting look very nearly normmal for most pics. There are some occasionally that look distinctly non-normal, but for the most part I think we can assume that in general a normal distribution is closely approximated. I'll go one further, and hypothesize that the variance is nearly the same for all.
Since this distribution is symmetrical, if we assume the "average" photo gets a 5.5, then the best and worst should be equidistant above and below that level. The fact that the tails of the distributions get truncated at 10 (or 1) is of no consequence, since it happens equally at both the high and low end.
I conclude that the voting here is incredibly impartial and that an average photo (average for dpchallenge, that is, which I think is significantly above average) does in fact get approx. a 5.5
Only a true geek would find that intensely interesting... oh well, once a geek, always a geek.
Message edited by author 2003-02-25 23:46:44. |
|
|
|
02/25/2003 10:26:33 PM · #5 |
Ah, I see.
Message edited by author 2003-02-25 22:26:57. |
|
|
|
02/26/2003 03:55:36 AM · #6 |
Coincidentally, I have been looking at the last few open challenges to find the average (arithmetic mean) score, just to see where I scored relative to it. The results are:
Road Signs Revisited: 5.354
Square: 5.178
Cliché: 5.268
Waldo: 4.902
Yellow: 5.156
As they are all lower than the 5.5 mid-point of the scoring range, it suggests that, on average, we are a fairly critical group of voters.
Kevin the geek
p.s. Can we have the average score on the challenge archive pages along with the min and max?
|
|
|
|
02/26/2003 04:10:02 AM · #7 |
Originally posted by paynekj:
As they are all lower than the 5.5 mid-point of the scoring range, it suggests that, on average, we are a fairly critical group of voters. |
I actually think that's because a large portion of voters don't understand that 5.5 is the average score. They think 5 is average, so they give that to photos they think deserve it. I wouldn't be surprised to find that voting as a whole is skewed slightly by this mistake. |
|
|
|
02/26/2003 04:21:44 AM · #8 |
| this 5.5 could be attributed to the 'i won't give ones' philosophy, which skews the results from 5.0 accordingly. i like to look at my mean score as the best indication of my photo's worth. |
|
|
|
02/26/2003 05:18:04 AM · #9 |
Originally posted by lisae:
I actually think that's because a large portion of voters don't understand that 5.5 is the average score. They think 5 is average, so they give that to photos they think deserve it. I wouldn't be surprised to find that voting as a whole is skewed slightly by this mistake. |
But we can't give a 5.5 vote!
I use 5 for an 'average' photo and I like the fact that it gives me more range in the 'above average' scale. |
|
|
|
02/26/2003 09:30:15 PM · #10 |
What does the numerical average have to do with whether or not a photo is good ?
"The fact that the tails of the distributions get truncated at 10 (or 1) is of no consequence, since it happens equally at both the high and low end."
Equally at both the high and low end ? I think there is an error in your process here. |
|
|
|
02/26/2003 09:59:33 PM · #11 |
Originally posted by David Ey:
What does the numerical average have to do with whether or not a photo is good ?
"The fact that the tails of the distributions get truncated at 10 (or 1) is of no consequence, since it happens equally at both the high and low end."
Equally at both the high and low end ? I think there is an error in your process here. |
David,
What I meant is that, overall, roughly as many distributions will get truncated at the top as will at the bottom. In the grand sceme of voting it all should average out. Roughly.
I noticed the skew from 5.5 toward 5.0 as well, and I believe the correct reasons have been proposed. I too tend to think of an "average" shot as a "5".
Last night I actually took the top 10 photos from each of the last 10 challenges, entered the totals for each score (1-10), and calculated the mean & variance. Yes, this meant I had to enter 1000 numbers. The results only partially supported my above expressed opinion that the variance is roughly constant. While this turns out to be very roughly true, certain photos have anomalously high voting variances, in other words they generated significantly more voter dissention!
Hmmm, a way to track the "controversy index" of submissions?
Message edited by author 2003-02-26 22:00:36. |
|
|
|
02/26/2003 10:05:01 PM · #12 |
| We could offer a 0-10 scale and let five be the average. We could also offer an "abstain" choice for those who have viewed the photo but are unable/unwilling to assign a numerical value. |
|
|
|
02/27/2003 05:49:28 AM · #13 |
Being v new here, I was surprised that there was a points system, rather than a prizes one (i.e. where you vote for your favourite image from the entries, or your favourite three, or whatever). The critique thing I like; though I don't know how much it gets used.
E my site |
|
|
|
02/27/2003 06:07:21 AM · #14 |
Let's compare my personal subjective scale with mathematics...
My scale:
1-Horrible beyond imagination
2-Bad "snapshot"
3-"Snapshot"
4-"Good snapshot"
5-Quality image
6-High quality image
7-Professional level
8-Work of Art
9-Week's Top three
10-Greatness
In my subjective scale I would agree with the mathematicians that the average submission would be in the range "quality" to "high quality" image. I think 5-6 are really good pictures.
I would also agree that average week winners in the 7-8 range are definitely "professional" to "Work of Art" quality.
I am new to the group so I haven't had an opportunity to give a 1 or a 10 yet, but I've been tempted at both ends.
My top three for the week always get a 9 after going through all images at least twice. That way I can compare my judgement to that of the group.
Judging photography is subject. We give a high score to something we like and a low one to those we don't. Some have a more descriminating eye than others, but as a group I think you are right on.
|
|
Home -
Challenges -
Community -
League -
Photos -
Cameras -
Lenses -
Learn -
Help -
Terms of Use -
Privacy -
Top ^
DPChallenge, and website content and design, Copyright © 2001-2025 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 07/23/2025 10:33:14 PM EDT.
